1. Field of the Invention
The present invention relates in general to computers, and more particularly to methods, computer systems, and computer program products for replicating data, especially in a deduplication system.
2. Description of the Related Art
Replication models usually provide either synchronous or asynchronous with regards to how and when replication is performed. In the synchronous model, data is replicated to the target as soon as possible after the data is stored to the source. Some models may do this as part of transaction processing such that the data is not considered successful until it is written to both sites while others may choose to allow transaction/operation completion at the source with the replication happening immediately after that. In either case, this is synchronous or near-synchronous in terms of what is provided and the expectations being achieved for the end-user. On the other hand, asynchronous allows for some mechanism to schedule replication. In this case, the scheduled replication is expected to be within some sort of timely manner to meet recovery point objectives while balancing those with bandwidth/capacity available for the replication.
In some storage management systems, server to server replication is provided as an asynchronous replication capability that is scheduled and managed through administrative oversight. While there are many benefits to replication, this may still represent a bandwidth and capacity issue when replication is performed. As an example, if the replication needs to process hundreds of thousands or millions of files within a defined replication window, considerable system and network resources may be required.